Starting from the simulated values obtained in the previous SIMEX phase, it computes averages across all simulations for each error rate and fits a function of the estimate in terms of error rate. From this function it extrapolates the value corresponding to the null error rate. The fitted function can be linear, quadratic, cubic or non linear.
extrapolation(results, lambda, lambda0, estimate0, fitting.method, B, parameter)A dataset with $100$ or B values for each error rate given by the simulation part of the algorithm.
A numerical vector containing the error rates.
The initial error rate.
The initial error prone estimate.
A string or a vector of strings containig the fitting methods for the function. It can be: 'line', 'quad', 'nonl' or 'cubi'.
The number of simulation for each error rate.
A string containg the parameter of interest P-SIMEX is performed on. It can be either 'inbreeding' or 'heritability'.
For inbreeding a list:
Inbreeding depression extrapolated error free value
Standard error of the error free value: regression component
Standard error of the error free value: sampling error component
Total variance of the error free value
The AIC of the fitted function
Heritability extrapolated error free value
Standard error of the error free value: regression component
Standard error of the error free value: sampling error component
Total variance of the error free value
Extrapolated value for the additive genetic variance
Extrapolated value for the environmental variance
The AIC of the fitted function